Personal Recommendation Based On The Weighted Association Rules And Browser Behavior

Along with the development and popularization of the information superhighway, the people were surround in ocean information. The information on the internet resources presents the index number inflation,is the letter source of the sea quantity, and the organization of its information is different,diverse and distribute. As a result, can provide the valid information recommendation for the users, help the user find out need of have the characteristic recommendation system of be worth the information to acquire the extensive concern in the Web information index fields, and also got the extensive application in personalize information service system.But, along with the personalize service system scale more and more big, the personalize recommendation system also faces a series of challenge.Such as:The valid recommendation design problem of the calculate way, real hour and the equilibrium problem of the recommendation quantity, the recommendation system structure problem of the system, recommendation result explain problem etc..This paper aims at the key techniques, such as the recommendation calculate way design within characteristic recommendation system and the recommendation system system structure...etc., carrying on the more thorough and beneficial quest and researches. First, we did the certain research to the personalize service recommendation system structure, propsing a kind of can distinguish analyse the long-term interest of user to provide the information recommendation the new personalize recommendation model with interest in the current future. Then, did the beneficial quest to the balance between real hour and the recommendation quantity of the recommendation system, not only propose that by catching the user of browse the behavior and obtain the user current interest, formationing user the current interest view , Combine to provide for the user in time valid of in the curren times the personalize recommendation method of the interest information recommendation; And propose a kind of by off-line add the weighted associate rule to discover the item of connection with acquire the long-term concern item of user, thus for the lately personalize recommendation that the user provide the accurate and valid customer long-term concern item information recommendation method. At provide the interested in information recommendation for the customer in the current times, because the behavior of user to browse web page can responds the user to browse interest from a certain degree, so we carry on succeed in catching to the user browse the behavior first, and adopt diverse return to return the metered relation that the method calculation the fixed amount relation of customer browses of behavior and interest in the web page ,and combines the web pages the tree( WPCT), formationing the current interest view ( CIV) of user, and make use of the CIV provide the based on conent collaborative filtering information recommendation for user. At provide the information recommendation of the long-term concern item for the user, we adopt first a kind of weighted associate rule calculate way lately to discover the item of associate, then prospsed the choice concern, combine the confidence of rules to get the recommendation degree rule with connection it.Make use of the recommendation degree to carry on the recommendation of long-term interest for the customer. Finally, we did the more overall emulation to experiment to the improvement method of this paper proposed. Analyze the enunciation, the method in this paper can improve the quantity of the information recommendation and the real hour of the recommendations availably, and the calculate way is reliable.The paper research work has the good and academic reference value and the good of applied value in personalize service recommend the system...